Page forty is imaginary. So is the note on it, and so are the patient, the consultant and the hospital. I say this at the start because the title sounds like a memory, and it is not one. It is a device, a way of holding a common kind of failure still long enough to take it apart. The kind of failure is real. Among the weaknesses of paper medicine set out in the flagship essay on this site, The Primary Care Office After the EHR, was a single line: a consultant’s recommendation could be buried dozens of pages deep. This essay follows one such recommendation, from the moment it is written to the moment, if it ever comes, when someone acts on it.
Here is the hypothetical case. A woman is admitted to hospital with an illness that has nothing to do with what follows; call it a chest infection. Her case file is a chronological bundle, every new sheet added at the back, and by the end of her stay it runs to some sixty pages. On her third day the treating team, puzzled by something on a scan, asks another department for an opinion. The consultant comes, examines her, reads the images, and writes a careful note on page forty. Most of it concerns the question the team asked, and answers it reassuringly. Near the end, almost in passing, the consultant mentions a second, incidental finding, small and probably harmless, and recommends a repeat scan in three months to make sure it has not changed. The patient goes home on her sixth day. Her discharge papers describe the infection and its treatment. They do not mention the scan. Three months pass, and then six. Nobody arranges it, and nobody can be said to have forgotten it, because nobody who would have arranged it ever knew it was there.
I have made the recommendation clinically vague on purpose. The essay is not about a particular finding or a particular test, and it offers no advice about either. It is about the distance between a sentence in a record and an act in the world, and it argues that the distance is crossed in at least four steps, each of which can fail on its own. A recommendation must exist: written, signed, dated and clear. It must be discoverable by the person who needs it. It must be accepted, or deliberately declined, by someone with the authority to decide. And it must be carried out, which means that a particular person orders a particular thing and checks that it happened. Page forty is a story about the first step succeeding and the other three failing in silence.
A word about the title’s other noun. In British and Indian usage a consultant is a senior specialist, the person at the top of a hospital team. In American usage the word more often names any specialist whose opinion is sought. The two meanings converge in the act this essay is about: someone is asked a question, and writes an answer into somebody else’s record.
A Note That Exists #
Existence sounds like the easy part. The note is on page forty; it exists. But existence in a record has conditions, and a careful reader checks them before relying on anything. A note must be attributable: the reader must know who wrote it, in what role, with what authority to advise. It must be dated, and in a hospital, where things change by the hour, timed. It must be legible, a condition whose history is told in What Paper Charts Did Terribly. And it must be complete in a particular sense: it must answer the question that was asked, and make plain where it goes beyond it.
That last condition fails more often than one might expect, and it was measured more than forty years ago. In 1983 Lee, Pappius and Goldman studied the medical consultations requested in their hospital over a week early in the academic year and a week late in it. Even by what they called very liberal definitions, the physician asking for the consultation and the consultant disagreed completely, on both the reason for the consultation and the principal clinical issue, in 22 of 156 cases, about one in seven. Consultants were twice as likely as the requesting physicians to rate their consultations as crucial to management, and the gap was widest precisely when the two disagreed about why the consultation had been called. A note can exist, be signed, dated and legible, and still be an answer to a different question from the one asked.
Goldman, Lee and Rudd drew practical conclusions in a short paper the same year, which they called “Ten commandments for effective consultations.” The consultant, they wrote, should determine the question being asked, establish its urgency, gather primary data, communicate as briefly as appropriate, make specific recommendations, provide contingency plans, understand his own role, offer educational information, communicate recommendations directly to the requesting physician, and provide appropriate follow-up. Read with page forty in mind, the list is remarkable for how much of it concerns not what the note says but whether it will travel. Brevity is a condition of being found. Specificity is a condition of being acted on. Direct communication admits that the page alone may not be enough. Follow-up admits that the consultant’s work is not finished when the note is signed. Half the commandments are instructions for getting a recommendation off the page.
Our hypothetical consultant obeyed most of them and broke two. The incidental finding was mentioned near the end of a long note devoted to another question, so that the recommendation was neither brief in its setting nor set apart from it; and nothing suggests that it was communicated directly to anyone. In the paper world there was a further hazard to existence, which electronic records have reduced but not removed: the opinion given aloud on a ward round and never written down, recorded afterwards, if at all, by a junior doctor’s line that says seen by the consultant, discussed. Such a line proves that a conversation took place. It does not say what was decided in it.
The Second Question #
The recommendation on page forty has a feature that makes it unusually vulnerable, and it is worth isolating, because it explains why so many lost recommendations share it. It answers a question nobody asked.
A consultation is a conversation with two parties. The treating team asks; the consultant answers; the answer returns to the asker, who is waiting for it and will read it with the question in mind. A recommendation that answers the question asked therefore has a natural owner before it is even written: the person who asked. Whatever else goes wrong, someone is expecting it. But consultants, being conscientious, often see more than they were asked to see. They notice a second thing on the scan, a medication that interacts with another, a finding in the blood that bears on nothing the team was worried about. They write it down, because leaving it out would be worse, and in writing it down they create a recommendation with no asker. Nobody is waiting for it. It arrives at the treating team as an aside to an answer they have already received, and the answer is what they read.
This is the orphan’s problem, and it explains much of page forty’s fate. The reassuring answer to the team’s question was found at once, because it was looked for. The recommendation about the incidental finding was not found, because nothing in anyone’s mind was shaped to receive it. It had existence without expectation, and in a busy hospital expectation is what turns a sentence into a task.
The consultant faces a real dilemma here, and it should not be caricatured. To say nothing about an incidental finding is to leave a possibly important fact unrecorded. To record it is to create an obligation that may be nobody’s: the consultant does not own the patient’s care, the team did not ask, and the patient may not know. Some consultants resolve the dilemma by writing the incidental recommendation separately and prominently, under its own heading, and by saying to whom it is addressed and who should arrange it. That small act of drafting converts an aside into an address. It is one of Goldman’s commandments applied to the question the consultant gave himself. An answer to an unasked question needs an asker invented for it, and the only person who can invent one is the consultant who wrote it.
Where a Reader Looks #
Edgar Allan Poe wrote the most searching story I know about failing to find a document that is in plain sight. In “The Purloined Letter” (1844), a compromising letter has been stolen from a royal apartment by a government minister, and the Prefect of the Paris police has searched the minister’s house with heroic thoroughness. He describes the search to the detective Dupin with pride. His men examined the rungs of every chair with a powerful microscope, looking for the dust a gimlet would leave. They probed the beds and the curtains and the carpets. Then they “divided its entire surface into compartments, which we numbered, so that none might be missed,” and scrutinised every square inch, and the two houses adjoining. They found nothing, and the Prefect has concluded that the letter is not on the premises.
Dupin’s diagnosis is that the Prefect’s method was flawless and his model wrong. The police, he says, “consider only their own ideas of ingenuity; and, in searching for anything hidden, advert only to the modes in which they would have hidden it.” They looked everywhere a letter could be hidden and nowhere a letter could simply be. When Dupin calls on the minister, wearing green spectacles so that his eyes can wander unobserved, he finds the letter in a cheap card-rack hanging from the mantelpiece, soiled and crumpled and torn nearly in two, as if about to be thrown away. It has been turned and re-sealed and re-addressed in a different hand, so that it looks nothing like the letter the Prefect was given a minute description of. It escaped the search not by being hidden but by no longer matching the description the searchers carried in their heads.
Every reader of a medical record carries such a description. She knows, or thinks she knows, where recommendations live: in the section for consultations, at the end of a specialist’s note under a heading that says recommendations, in the discharge summary, in a letter addressed to her. The recommendation on page forty lives in none of these places in the expected form. It is a sentence near the end of a note about something else, phrased as a suggestion, sitting where the Prefect would never look because it was never hidden. A reader can turn every page of a sixty-page file with care, as the Prefect turned every chair, and still pass it, because her attention is organised by the question she brought, and the question she brought was about the chest infection.
Dupin has a second observation that bears on records more uncomfortably still. In a game played on a map, he says, novices challenge their opponents to find names printed in the smallest letters, while experts choose names that stretch in large characters from one side of the map to the other, because these, like the over-large signs and placards of the street, “escape observation by dint of being excessively obvious.” A modern note can bury a recommendation in exactly this way: not in small print, but in a page so full of headings, bold type and repeated boilerplate that emphasis has stopped meaning anything. When everything is printed large, the eye stops reading the large print.
The page number in my title is therefore misleading in a useful way. Page forty sounds like an address, but a page number says almost nothing about whether a reader will find what is on the page. On paper, page forty of a chronological file sat in the middle, which was the worst place to be: the admitting team re-reading the start of the file would not reach it, and the discharging team writing from the end of the file would not go back to it. The remedy proposed more than half a century ago, the record organised by problem rather than by date, is part of the story told in the flagship essay, and it helped; a recommendation filed under the problem it concerns has a better chance of being seen by whoever is thinking about that problem. But an incidental finding is, by definition, not the problem anyone is thinking about.
Advice Is Not a Decision #
Before following the recommendation further, it is necessary to say what it is not, because a great deal of bad reasoning about missed recommendations begins by treating them as orders.
A consultant advises. The physician responsible for the patient decides, and decides with the patient. That division is not a formality; it exists because the consultant sees one part of the picture and the treating physician is supposed to see the whole of it. A recommendation can be right in itself and wrong for this patient: she may have other illnesses that change its balance of benefit and burden, or goals that make it unwelcome, or circumstances that make it impossible. It may have been conditional, a step to take only if something else turns out a certain way. Two consultants may have recommended incompatible things. Time may have overtaken it; a recommendation written on day three can be obsolete by day six. A system that carried out every recommendation automatically would not be safer. It would be a machine for converting partial views into complete actions.
Punjabi, like Hindi and Urdu, keeps this distinction in its everyday vocabulary. ਸਲਾਹ (salāh) is advice or counsel, and the pair ਸਲਾਹ-ਮਸ਼ਵਰਾ (salāh-mashvarā) names the whole business of conferring before a decision: the family that holds a salāh-mashvarā about a marriage or a sale of land is deliberating, not obeying. One takes advice, in Punjabi as in English, and taking it is not the same as following it. The word implies a second person, the one who listens to the counsel and then decides.
So the right question about page forty is not why was the recommendation not followed? but why was it never decided? A recommendation deliberately declined, with a reason written down and the patient informed, is a success of the system, even if the scan is never done. A recommendation that nobody accepted, nobody declined and nobody saw is the failure. The difference lies, whatever the outcome, in whether anyone ever held the decision in their hands.
Inheriting Someone Else’s Advice #
Suppose, against the odds, that page forty is found. Suppose the discharge summary is rewritten to include it, and the patient’s primary care physician reads, a week after she goes home, that a consultant she has never met recommends a repeat scan in three months for a finding she has not seen. What has been handed to her?
Not, I think, a simple task. She is being asked to accept responsibility for someone else’s judgment. She does not have the images. She may not have the consultant’s full note, only a line that summarises it. She does not know how firmly the consultant meant the recommendation, whether it was a matter of routine caution or of real concern, or whether it took account of the patient’s other conditions, which the primary physician may know far better than anyone in the hospital did. If she orders the scan, she will own its result, and whatever follows from that result. If she does not, she will own that decision too, on the strength of a sentence written by a stranger.
There is a quiet temptation in this position, and it runs in two directions. One is to carry out the recommendation without much thought, because the specialist must know best and declining it feels like taking a risk on someone else’s behalf. The other is to treat it as the hospital’s business, on the reasonable assumption that the consultant who made it will arrange it. Both are understandable; both leave the recommendation undecided in the sense this essay cares about. The first converts advice into an order without anyone having weighed it for this patient. The second returns it to page forty.
What the receiving physician needs, to decide well, is what the consultant’s note was supposed to supply in the first place: the reason, the degree of concern, the timing and what would change the plan. A recommendation that arrives with those things can be accepted or declined on its merits. A recommendation that arrives as a bare instruction can only be obeyed or ignored. The difference between those two pairs of verbs is the difference between a handoff of judgment and a handoff of paper.
Filed #
Platts’s great dictionary of Urdu and Hindi, compiled in the 1880s, records an idiom long familiar in the offices of northern India, and that Punjabi still uses as ਦਾਖ਼ਲ-ਦਫ਼ਤਰ (dākhil-daftar): literally, entered in the office. Platts gives its senses in an order I find almost comic in its candour: “Entered upon the record, placed on record; filed with the record; laid aside, or indefinitely postponed; struck off the file.” To place a matter on record and to shelve it indefinitely are, in this idiom, the same act. The revenue office had learned what hospitals keep relearning: filing a thing can be the most respectable way of not dealing with it.
Page forty is a recommendation that has been made dākhil-daftar in both senses at once. It was placed on record impeccably, and by that very act it was laid aside, because a recommendation that is only on the record belongs to nobody. The question of ownership is where most such failures actually happen, and it is the subject of another essay on this site, The Handoff, which follows a pending test result through a chain of reasonable assumptions that leave it without an owner. A recommendation suffers the same fate with one added disadvantage. A pending result is at least awaited; some system expects it and will deliver it somewhere. A recommendation for a future test is not awaited by anything. It exists only as a sentence, and a sentence has no way to notice that it has been ignored.
How common is the loss of such sentences across the walls between doctors? The best evidence concerns referrals and results rather than buried notes, but it points the same way. In a study published in 2000, Tejal Gandhi and colleagues reported surveys of primary care physicians and specialists at an academic medical centre in Boston about the referrals passing between them. Sixty-eight percent of specialists reported receiving no information from the referring physician before particular referral visits; four weeks after such visits, 25 percent of the primary care physicians had still received nothing from the specialist. A recommendation made in a specialist’s office and never sent is page forty with a postage problem.
The more revealing study looked at what happens even when a message plainly arrives. Hardeep Singh and colleagues examined critical alerts about abnormal imaging results sent through the electronic record at a large Veterans Affairs facility in 2007 and 2008. Of 1,196 alerts, 18.1 percent were not acknowledged, that is, not opened by the clinician they were sent to, within two weeks. That is troubling, but the more instructive finding came next. Timely follow-up was lacking in 7.7 percent of all alerts, and the rate was similar whether the alert had been acknowledged or not: 7.3 percent of acknowledged alerts against 9.7 percent of unacknowledged ones, a difference the study could not distinguish from chance. Opening a message and acting on it were different events, and the first did not reliably produce the second. Two other findings matter for page forty. Alerts sent to more than one clinician were more likely to go unacknowledged and more likely to lack follow-up, as though a responsibility shared were a responsibility halved. And when the radiologist had also spoken to someone directly, the odds of follow-up being missed were far lower. Twenty-six years after Goldman and his colleagues told consultants to communicate their recommendations directly, an observational study in a fully electronic system found a spoken word still doing what the record alone did not.
Tess’s Letter #
Thomas Hardy gave English fiction its most painful misdelivered document, and its painfulness lies in what happens after the failure is discovered. In chapter thirty-three of Tess of the d’Urbervilles (1891), Tess, unable to bring herself to tell Angel Clare about her past before their wedding, writes it all down in a letter of four pages and pushes it under the door of his room. In the days that follow he is as loving as ever, and she half believes he has read it and forgiven her. On the morning of the wedding, Hardy writes, her “perception that Angel’s bearing towards her still remained in no whit altered by her own communication rendered Tess guiltily doubtful if he could have received it.” She climbs to his room, stoops at the threshold, and sees under the edge of the carpet “the faint white margin of the envelope,” which in her haste she had thrust “beneath the carpet as well as beneath the door.” There it was, Hardy says, “sealed up, just as it had left her hands.” She destroys it. And then comes the sentence on which the novel turns: the misplaced letter seemed to prevent a confession, “but she knew in her conscience that it need not; there was still time.”
Two things in this scene seem to me exact about clinical communication. The first is how Tess learns that her message failed. Nobody tells her. She infers it from the receiver’s behaviour, which has not changed in the way it would have changed had the letter been read. In medicine, the equivalent signal is the absence of an expected event: no scan booked by the date it should have been, no reply to a referral, no new order after a recommendation. Such absences are evidence that a message was not received, but only to someone who was expecting the event and is watching for its absence. In the page-forty case nobody was watching, because nobody but the consultant knew there was anything to watch for, and the consultant had handed the matter to a page.
The second is what Tess does with her discovery. She had another channel available, the one she had been avoiding, which was to speak, and she tried it only once, lightly, on the landing, and let Angel put her off until they were settled. Hardy is clear that the lost letter had not forced her silence, and that Angel’s refusal reassured her more than it should have. A sender who learns that a written message has failed is not released by the failure; if the message matters, the failure obliges her to use the other channel, and to press it until it has been heard. For a consultant, the note is one channel, a direct word with the treating team is another, and the patient is a third.
That third channel needs careful handling. Telling the patient that a repeat scan is recommended, and why, and who will arrange it, is an independent path by which the recommendation can survive the loss of page forty. Patients often carry their own care across gaps that clinicians leave. But a patient in hospital is ill, often tired or in pain, frequently medicated, and told a great many things in a short time. The finding in this case had nothing to do with why she came in, and a remark that we’ll want to look at this again in a few months can easily be heard as reassurance rather than as a task. She may face costs, distances or a language barrier that make an appointment hard to keep even if she understood it. Informing her is redundancy, a second rope. It is not a transfer of responsibility from the people whose job it was to arrange the scan to the person who was being treated. The useful test is the one used in good discharge teaching everywhere: ask her to say back what will happen next, and who will do it. If she cannot, the second rope is not tied.
The Story Everyone Wants to Tell #
The page-forty story is usually told as a story about blame. Somebody should have read the note. Somebody should have put it in the discharge papers. Somebody should have arranged the scan. Each of these is true in a sense, and the telling is satisfying, because it locates the failure in a person and implies that a better person would have prevented it. It is also, very often, unfair, and the unfairness has been measured.
In 1991 Caplan, Posner and Cheney asked 112 practising anaesthesiologists to judge the appropriateness of care in 21 cases that had ended in harm. For each case they had written a matching version, identical in every respect except that the outcome was changed in severity, from temporary to permanent or the reverse. Reviewers, unaware of the design, saw one version or the other. In fifteen of the twenty-one pairs, the worse the outcome, the less appropriate the identical care was judged to be. Overall, the share of ratings calling the care appropriate fell by 31 percentage points when the outcome was made permanent. The reviewers were experienced physicians judging the conduct of their own specialty, and the outcome moved them anyway.
Page forty is exposed to exactly this distortion. Read after something has gone wrong, the recommendation seems to glow on its page, and the failure to find it looks like inattention. Read prospectively, by a team with sixty pages, five other patients and a question about a chest infection, it was a sentence among thousands. A fair account has to ask questions the blame story skips. Was the note somewhere a reasonable clinician, looking for what she had reason to look for, would have found it in the time she had? Was the recommendation still appropriate when the patient left? Would carrying it out have changed anything? My hypothetical leaves these unanswered on purpose, because the honest answer in a real case is often we cannot be sure, and because the design question does not depend on them. A more useful question than who failed to find this note is how the next one will be found, by someone whose job it is to find it, whether or not anyone is careless.
Search Is Not Follow-Up #
Electronic records have changed page forty’s chances considerably, and it is worth being precise about how. A note that once sat in the middle of a bundle of paper can now be found in seconds, from anywhere, by searching for a word. That is a real improvement, and it has surely rescued many recommendations that would once have been lost.
But search answers questions; it does not ask them. The recommendation is found in seconds by someone who already suspects it exists and knows roughly what words it uses. The physician who sees the patient three months later for a cough, with no reason to think a scan was ever recommended, will not search for it, any more than the Prefect would search a card-rack. And even when the note is found, finding it leaves the real work undone. Someone must judge whether the recommendation still applies, decide whether to accept it, order the test, make sure the patient knows and can attend, and look at the result when it comes. Retrieval solved the problem of the folder. It did not touch the problem of ownership, and Singh’s study suggests that even guaranteed delivery leaves a residue of messages that are opened and not acted on.
The tempting remedy is the alert: a flag that fires whenever a note contains a recommendation. It would fire constantly, because notes are full of suggestions, conditional plans and observations phrased as advice, and a clinician shown hundreds of such flags a week will learn to dismiss them as quickly as the system can display them. Singh’s finding that messages addressed to several people fared worse than messages addressed to one is a warning here too. An alert that goes to everyone gives each recipient a reason to think someone else will act. More signals do not make a recommendation more likely to be completed. A single owner does.
The flagship essay imagines software that can notice when a consultant’s recommendation never became an order, and I still think that is one of the more valuable things such software could do. The condition is the one this essay keeps arriving at. Noticing is the easy part. Unless the system also makes plain who has been asked to decide, and whether that person has said yes, it will produce one more record of the recommendation’s existence, and the recommendation will be dākhil-daftar again, this time electronically.
What Page One Should Know #
What would a record look like if it were built to carry recommendations to completion? I think it would treat a recommendation not as a sentence inside a note but as an object with a status of its own. When written, it would be open, and it would say plainly what is recommended, why, by when, and to whom it is addressed. It would remain open until someone with authority did one of four things to it: accepted it and took ownership, with a date by which it should be done; modified it, with a reason; declined it, with a reason and with the patient’s knowledge; or recorded it as completed, with the result. And the list of open recommendations, short, current and impossible to miss, would sit at the front of the record, on page one, where every clinician who opens the chart for any reason would see it.
None of this requires new technology so much as a change in what the record is asked to remember. Paper could have done it with a single sheet at the front of the file. Electronic records can do it better, if they are designed to. The principle is the one the commandments of 1983 pointed toward: a recommendation is not finished when it is written, and the writer’s work includes making sure it has somewhere to go.
The note will always be on some page. Records grow, and the forty-first page and the four-hundredth will follow the fortieth. The question is not where the note lives. It is whether its status lives on page one. Dupin found the purloined letter because he was clever enough to think like the man who had placed it. A record should not require that kind of cleverness of the tired people who read it. And when a recommendation has gone astray, the record should be able to say what Hardy’s narrator said of Tess’s letter, while it is still true: there is still time.
Evidence and Editorial Note #
Documented findings. Public biography is relied on only in general terms (a family physician; Punjabi as first language). Lee T, Pappius EM, Goldman L, “Impact of inter-physician communication on the effectiveness of medical consultations,” Am J Med 1983;74:106–112 (consultations over a week early and late in the academic year; complete disagreement on both the reason and the principal clinical issue in 22 of 156; consultants rating consultations crucial 35 versus 18 percent, especially when reasons differed): PubMed abstract, PMID 6849320. Goldman L, Lee T, Rudd P, “Ten commandments for effective consultations,” Arch Intern Med 1983;143:1753–1755 (the ten principles, paraphrased): PubMed abstract, PMID 6615097. Gandhi TK et al., “Communication breakdown in the outpatient referral process,” J Gen Intern Med 2000;15:626–631 (academic tertiary centre; 68 percent of specialists receiving no information before specific referral visits; 25 percent of primary care physicians with no information four weeks after): PubMed abstract, PMID 11029676. Singh H et al., “Timely follow-up of abnormal diagnostic imaging test results in an outpatient setting,” Arch Intern Med 2009;169:1578–1586 (Veterans Affairs facility, November 2007 to June 2008; 1,196 alerts; 18.1 percent unacknowledged at two weeks; timely follow-up lacking in 7.7 percent, 7.3 versus 9.7 percent for acknowledged and unacknowledged, P = .22; higher risk with dual alerts; lower with additional verbal communication by the radiologist, odds ratio 0.12): PubMed abstract, PMID 19786677. Caplan RA, Posner KL, Cheney FW, “Effect of outcome on physician judgments of appropriateness of care,” JAMA 1991;265:1957–1960 (112 anesthesiologists; 21 matched case pairs; inverse relation in 15; appropriate ratings down 31 percentage points when outcomes were made permanent): PubMed abstract, PMID 2008024. Edgar Allan Poe, “The Purloined Letter” (1844), quoted: The Works of Edgar Allan Poe, Raven Edition, vol. 2, Project Gutenberg text. Thomas Hardy, Tess of the d’Urbervilles (1891), chapter XXXIII, quoted: Project Gutenberg text. ਦਾਖ਼ਲ-ਦਫ਼ਤਰ (dākhil-daftar) and its quoted senses: Platts, A Dictionary of Urdu, Classical Hindi, and English (1884), entry dākhil, via the Digital Dictionaries of South Asia, with ordinary Punjabi usage. ਸਲਾਹ and ਸਲਾਹ-ਮਸ਼ਵਰਾ: ordinary Punjabi usage. The problem-oriented record as an organising remedy is referred to the flagship essay; British, Indian and American senses of “consultant” and the practice of teach-back at discharge: standard reference knowledge. Vendor representations. None. Electronic records, search and alerts are described in general terms; no product is named or assessed.
Recollection and tradition. No page, note, consultant, patient, hospital or incident from the author’s training or practice is described. The woman admitted with a chest infection, the sixty-page file, the note on page forty and its recommendation are a hypothetical illustration, stated as such at the outset; the clinical details are deliberately unspecific and are not advice. The remark on ward-round opinions recorded as “seen, discussed” is a general description of how paper records were sometimes kept, not a recollection.
Analysis and forecast. The four conditions of existence, discoverability, acceptance and execution; the readings of Poe, Hardy and the dictionary entry; the argument that a recommendation should be decided rather than automatically executed; the account of the incidental recommendation as an answer without an asker; the account of the receiving physician’s position; the reading of Singh’s findings on shared alerts and direct communication; the critique of the blame narrative; the doubt about alert-based remedies; and the proposal that recommendations carry a status listed at the front of the record are the author’s interpretation and preference as of November 7, 2026. They make no allegation about any real clinician, hospital or record. Independence. This essay is not sponsored by any institution, company, health system or technology developer. It is personal reflection, not medical advice; readers seeking care should contact their own physician’s office.
About the Author #
Kanwar Partap Singh Gill, MD, traces his family roots to Khadur Sahib in Punjab’s Majha region and grew up in Amritsar. He attended Spring Dale School and completed pre-medical studies at Khalsa College in 1994–1996. He earned his MBBS from Sri Guru Ram Das Institute of Medical Sciences and Research in Amritsar in 2002.
He completed PGY-1 at the University of Arkansas for Medical Sciences in Little Rock in 2007–2008, and PGY-2 and PGY-3 with UCSF Fresno at Community Regional Medical Center from 2008 through 2010. He has practiced medicine in California since 2010.
His career has crossed four distinct information eras: predominantly paper-based medicine, early and fragmented electronic records, the integrated-EHR era, and the emerging age of generative and agentic artificial intelligence.